The present invention relates to lasers and non-linear frequency conversion techniques, and particularly to a technique for simultaneously producing two or more wavelengths which are converted to a third wavelength using intracavity sum frequency generation.
Lasers are well-known devices that produce monochromatic optical radiation. Lasers contain an optical gain medium in a resonator. The laser resonator is composed of at least two reflective elements that provide optical feedback. These reflective elements define the laser resonator cavity, which is the volume contained along the reflective feedback path within the resonator.
Extremely narrow optical bandwidths can be obtained, and laser emission with a spectral purity approaching one part in 10.sup.14 has been demonstrated. However, the single emission wavelength, which for many applications is an important advantage of a laser, can also be a serious drawback. For example, numerous applications require specific, discrete wavelengths. In many instances there is no laser transmitter that operates at that specific wavelength, so that tunable lasers are highly suited to this type of application.
Tunable lasers can be either discretely tunable, by which is meant two or more specific wavelengths can be obtained from a given laser, or continuously tunable, by which is meant that over a certain band (or preselected range) of wavelengths any arbitrarily selected wavelength may be obtained. In addition, with continuously tunable lasers the emission bandwidth can be made extremely narrow to provide the possibility of obtaining truly monochromatic optical radiation from a given laser over a wide wavelength band.
Dye lasers are one example of a class of lasers that are continuously tunable. The range of tunability is typically less than 100 nm, and dye lasers are well-known to produce extremely narrow output bandwidths. Dye lasers operate in the wavelength range of approximately 400 nm to 750 nm. Dyes operating beyond 800 nm are usually unstable and are, in general, impractical. To cover the wide bandwidth range of 400 nm to 750 nm, typically five to six different dyes are needed.
While dye lasers and other continuously tunable lasers can be useful for applications requiring a discrete wavelength, other applications may simultaneously require more than one wavelength. Such applications include resonant two-photon spectroscopy, for example, or sum frequency generation (SFG). For SFG a non-linear optical crystal is utilized to produce an emission at a wavelength by summing the frequencies of radiation at two different wavelengths. If the two fundamental wavelengths are represented by .lambda..sub.1 and .lambda..sub.2, then the sum frequency wavelength, .lambda..sub.3, is given by the expression: ##EQU1## From Equation 1 it can be seen that .lambda..sub.3 is smaller than either .lambda..sub.1 or .lambda..sub.2. This technique is used commonly to obtain visible emission from infrared (IR) emitting lasers.
There are several types of lasers which have been demonstrated to simultaneously produce two wavelengths. Early demonstrations of multifrequency devices concentrated on pulsed dye lasers. See for example, H. S. Pilloff, "Simultaneous Two-wavelength Selection in the N.sub.2 Laser Pumped Dye Laser," Applied Physics Letters, vol. 21, pp. 339-340, 1972; C. Wu and J. R. Lombardi, "Simultaneous Two-frequency Oscillation in a Dye Laser System," Optical Communications, vol 7, pp 233-236, 1973; and H. Lotem and R. T. Lynch, Jr., "Double Wavelength Laser," Applied Physics Letters, vol. 27, pp. 344-346, 1975.
The techniques used to demonstrate multifrequency dye lasers were generally oriented toward pulsed laser systems. These techniques typically use inefficient means to separate and tune the output wavelengths and are not suitable for continuous wave (cw) operation. More recently, a titanium-doped sapphire (Ti:sapphire) laser was demonstrated to operate cw multifrequency. This laser uses a tunable solid state laser gain element which operates over the wavelength range of about 680 nm to about 1.1 .mu. and is ideally suited for a number of applications that previously had used dye lasers. As mentioned above, some laser dyes that emit in the near-IR range of about 800 nm to 1 .mu., tend to be unstable. The Ti:sapphire laser is a much more practical way of achieving cw tunable operation in this wavelength range. The Ti:sapphire gain element is a crystalline material that is typically shaped as a cylindrical laser rod with Brewster angle end faces.
The demonstration of a doubly resonant cw Ti:sapphire laser was recently reported in the literature. See for example, R. Scheps and J. F. Myers, "Doubly Resonant Ti:sapphire Laser," IEEE Photonics Technology Letters, vol 4, pp. 1-3, 1992. This same gain element, Ti:sapphire, had also previously been demonstrated to operate simultaneously at two wavelengths ("multifrequency") in a pulsed mode similar to the type of operation that had been previously obtained in dye lasers. See C. Kruglik, P. N. Nazarenko, N. V. Okldnikov, F. A. Skripko and A. A. Stavrov, "Autonomous Tunable Multifrequency Near-IR Laser," Atmospheric Optics, vol. 2, pp. 729-734, 1989; and S. G. Bartoshevich, I. V. Mikhnyuk, F. A. Skripko and I. G. Tarazevich, "Efficient Difference Frequency Oscillator Based on a Ti:sapphire Laser," IEEE Journal of Quantum Electronics, vol. 27, pp. 2234-2237, 1991.
Multifrequency operation of a tunable laser is desirable when there is independent wavelength and bandwidth control of each output wavelength. Such a device can be used more readily for numerous applications. U.S. Pat. No. 4,287,486 entitled "Laser Resonator Cavities with Wavelength Tuning Arrangements" by Ali Javan discusses various means of using tunable gain media to obtain multifrequency operation. All wavelengths in the Javan laser can emerge collinearly. Independent control of wavelength and spatial separation between wavelengths is provided. There are several apparent limitations in Javan's patent related to the wavelength tunability technique employed and to the spatial transverse mode control of the multifrequency laser output.
Regardless of the technique used to achieve simultaneous multifrequency operation in the pulsed or cw lasers, or discretely tunable or continuously tunable lasers discussed above, all techniques limit emission to a preselected range of wavelengths over which the gain material can produce optical gain. The gain of a given laser material at a specific wavelength is determined by the stimulated emission cross-section, .sigma.. Numerous factors determine the spectral dependence of the stimulated emission cross-section. For a laser to operate at a given wavelength, the gain experienced at that wavelength must exceed the sum of all losses incurred. Losses are due to such factors as transmission through the resonator (output coupling), scattering and absorption.
The Ti:sapphire laser operating between 680 nm and 1.1 .mu.has one of the largest tuning ranges of any known laser material. Other gain materials such as Cr.sup.3+ -doped solid state materials and dye lasers have much narrower tuning ranges, typically on the order of 80 nm to 200 nm. If simultaneous dual wavelength operation is desired at two wavelengths for which no single gain material demonstrates gain, the techniques previously described and demonstrated cannot be used. A different type of laser is required.
Non-linear optical conversion is commonly used to produce visible radiation from solid state lasers operating in the near infrared. The most common non-linear optical conversion technique is a process called second harmonic generation, or doubling. In the doubling process the laser output is directed through a non-linear material, generally a crystal. The laser beam emerging from the non-linear crystal is at one-half the wavelength of the initial laser beam (i.e., the optical frequency is doubled). The initial laser wavelength is referred to as the "fundamental wavelength" and the shorter wavelength is called the "second harmonic".
An important parameter for efficient non-linear conversion is phase matching. Optimum conversion from the fundamental wavelength to the second harmonic wavelength will occur when the wave vector mismatch between the fundamental wave and the second harmonic wave is zero. This condition is termed "phase matching". Phase matching may be achieved in an anisotropic crystal by a suitable choice of the direction of propagation relative to the crystalline axes.
Two different types of second harmonic generation (SHG) can occur in non-linear crystals. The second harmonic process can be thought of as mixing two waves of identical wavelength to produce a third wave at one-half the wavelength. In this context, Type I SHG refers to the process where the two fundamental waves have the same polarization. Type II SHG occurs when the two fundamental waves have orthogonal polarization with respect to each other.
Phase matching is achieved as a result of the dispersion of the non-linear crystalline material. Dispersion refers to the dependence of the refractive index of a given material on wavelength. Phase matching is achieved in second harmonic generation when the refractive index at the fundamental wavelength is equal to the refractive index at the second harmonic wavelength. Because of dispersion, the refractive indices at the two wavelengths can be equal if the material is birefringent. That is, the crystal must have a different refractive index for the ordinary and extraordinary waves. In this case a propagation direction may be chosen with respect to the crystallographic axes where for a given fundamental wavelength the refractive indices for the second harmonic and the fundamental wavelengths are identical.
The conditions for phase matching depend specifically on the desired non-linear operation. For example, second harmonic generation of the Nd:YAG laser wavelength from 1.064 .mu. to 532 nm requires that the refractive index of the non-linear crystal at 1.064 .mu. and the refractive index at 532 nm be identical. In such a case, the phase relationship between the fundamental wavelength and the generated second harmonic wavelength remain unchanged as the two waves propagate along the length of the non-linear crystal.
When phase matched second harmonic generation is achieved by propagating the fundamental wavelength along a direction different from a principal axis of a birefringent crystal, it is termed "critical phase matching". When critically phase matched second harmonic generation is obtained with a focused beam, there is a phase mismatch of the wave vector for small deviations from the phase matched direction due to the finite divergence of the beam. However, since the efficiency of the non-linear conversion process is a function of the power density within the non-linear crystal, focusing is generally desirable to achieve high conversion efficiency.
When the phase matching angle is 90.degree. for a particular non-linear process in a given material, it is termed "non-critical phase matching" (NCPM). In such a case, effects of beam divergence vanish. That is to say, a strongly focused beam in an NCPM crystal does not have the phase mismatch problems as is evident in critical phase matching. In addition, the walk-off angle, which is the angle of energy flow of the fundamental relative to the second harmonic beam, is zero. This allows the two beams (the fundamental and second harmonic) to propagate collinearly within the crystal.
NCPM is therefore a desirable and potentially highly efficient type of phase matching. One means by which NCPM can be obtained is by adjusting the temperature of the non-linear crystal to the point where the refractive index of the fundamental wavelength equals that of the second harmonic wavelength for a particular propagation direction with respect to the crystallographic axes. NCPM can also be achieved at room temperature for certain non-linear materials and fundamental wavelengths. Room temperature NCPM has the advantage of simplicity.
Second harmonic generation is a special case of a more general non-linear optical conversion process known as sum frequency generation. In second harmonic generation, two optical waves of the same wavelength are combined to produce a single wave of a wavelength one-half the original fundamental wavelength. In sum frequency generation, two fundamental waves of different wavelengths are combined to produce a third wavelength. The wavelength produced by sum frequency generation is determined by equation (1). Second harmonic generation is a degenerate case of sum frequency generation since for second harmonic generation .lambda..sub.1 =.lambda..sub.2. The fundamental principles of non-linear optics summarized briefly above are well known and are discussed in detail in the literature. See for example, G. D. Boyd and D. A. Kleinman, Journal of Applied Physics, vol. 39, p. 3597, 1968.
In principle sum frequency generation (SFG) can be more efficient than second harmonic generation (SHG) under certain conditions. If a wavelength .lambda..sub.3 is desired, it can be obtained by SFG using a particular .lambda..sub.1 and .lambda..sub.2 labeled .lambda..sub.o1 and .lambda..sub.o2. Obtaining .lambda..sub.3 by SHG requires a fundamental wavelength .lambda..sub.1 (equal to .lambda..sub.2) equal to 2.lambda..sub.3. If the wavelength 2.lambda..sub.3 is not near the peak wavelength for a given laser, then the efficiency of producing 2.lambda..sub.3 may be low. It is often the case that 2.lambda..sub.3 is not near the peak emission wavelength or cannot be produced efficiently.
In addition, it is difficult to operate a laser at high power when it is tuned substantially off the peak output wavelength. If the fundamental wavelength 2.lambda..sub.3 is obtained from a laser for which the stimulated emission cross-section at 2.lambda..sub.3 is substantially lower than the peak stimulated emission cross-section, the wavelength corresponding to the peak stimulated emission cross-section may be emitted even though the laser is adjusted to operate at 2.lambda..sub.3.
On the other hand, using SFG it may be possible to identify two lasers such that .lambda..sub.o1 and .lambda..sub.o2 represent wavelengths for each laser which are at, or close to, the wavelengths for which the spectral dependence of the stimulated emission cross-section has a peak. Then the production of .lambda..sub.3 by summing .lambda..sub.o1 +.lambda..sub.o2 has significant advantages compared to the case where one must double the frequency at wavelength 2.lambda..sub.3, for which the stimulated emission cross-section might be substantially lower than that at the peak.
In general, sum frequency generation requires two lasers. This can introduce complexity in terms of alignment since beam spot sizes in the non-linear crystal must be quite small for efficient sum frequency generation.
An example of efficient sum frequency generation is given by the summation of the wavelengths 808 nm and 1.064 .mu. to produce 459 nm. The non-linear crystal KTiOPO.sub.4 (KTP) is non-critically phase matched at room temperature for sum frequency generation at these two wavelengths. See K. Kato, IEEE J. Quantum Electronics, vol. QE-24, p. 3, 1988. Since 1,064 .mu. is the peak wavelength for Nd:YAG lasers and 808 nm is near the peak wavelength of several Cr.sup.3+ -doped solid state lasers as well as AlGaAs laser diodes and Ti:sapphire lasers, the sum frequency generation process can proceed quite efficiently using these two wavelengths to produce 459 nm.
On the other hand, achieving 459 nm through SHG requires that a laser be designed to produce 918 nm output. This wavelength is difficult to generate efficiently as only a few tunable lasers cover this wavelength range. In addition, 918 nm is not near the peak for the stimulated emission cross-section of any efficient, scalable laser. Furthermore, doubling 918 nm generally uses non-linear crystalline materials that are not as mature as KTP, such as KNbO.sub.3, which is also less robust than KTP. Finally, it is difficult to obtain a high quality crystal which is non-critically phase matched at room temperature for second harmonic generation of 918 nm emission to produce output at 459 nm.
Typically, as mentioned above, SFG requires two different laser sources. Since the efficiency of the sum frequency generation process depends upon the power density (power per unit area) within the non-linear optical crystal, extremely small focused spot sizes within the non-linear sum frequency generating crystal are desirable. Using two different laser sources generally leads to problems involving the alignment of the beams to the high degree of accuracy required by these small spot sizes within the non-linear crystal. In addition, when using two separate laser sources, there are inefficiencies that result from spatial mismatching at the beam focus from the two lasers in terms of size, shape and intensity distribution.
It should be noted that for cw sum frequency generation the use of the non-linear SFG crystal within a resonator cavity ("intracavity") is desirable, since high power densities produce efficient SFG output. Therefore, the non-linear crystal used for cw SFG usually receives focused light within a laser resonator cavity to take advantage of the high circulating optical flux. The intracavity optical power is much higher than the power outside of the cavity. Note also that when a single laser resonator resonates both fundamental wavelengths .lambda..sub.1 and .lambda..sub.2 used for SFG, there should be no elements contained within the cavity that are strongly absorbing at either wavelength. Although good efficiency dictates the application of intracavity sum frequency generation to take advantage of the high circulating power at the fundamental wavelengths .lambda..sub.1 and .lambda..sub.2 for cw operation, for pulsed operation the non-linear optical crystal is placed external to the laser resonator cavity to obtain good sum frequency generation efficiency. However, in all cases small spot sizes within the laser crystal are desirable to enhance the conversion efficiency by increasing the power density. The maximum sustainable power density is determined by laser-induced optical damage to the nonlinear crystal or optical coatings on the crystal faces, as well as by thermal effects which created phase mismatch. Efficient SFG requires good alignment and good spatial mode matching between the beams representing the two fundamental wavelengths, .lambda..sub.1 and .lambda..sub.2.
Many of these requirements can be readily satisfied if both .lambda..sub.1 and .lambda..sub.2 are produced simultaneously by a single laser. In general, in order to achieve simultaneous operation at both wavelengths, the net round-trip gain at each wavelength must be comparable within the laser resonator cavity. However, the spectral dependence of the stimulated emission coefficient for a given laser material (which determines the intrinsic gain of the material) makes it unlikely that the net gain would be the same at both fundamental wavelengths for the SFG process, particularly if the two wavelengths are very different.
Thus, in accordance with this inventive concept a continuing need has been found in the state of the art for a laser for producing two wavelengths simultaneously to be used for sum frequency generation or other applications. Such a laser is useful where the separation between the two wavelengths is arbitrarily large, where it is desirable to scale the laser emission to higher output power, where efficient sum frequency generation insensitive to alignment is desired, and where the spatial modes at both wavelengths are approximately identical. In addition, such a laser can be pumped by a single pump source and can be used for intracavity sum frequency generation in a cw or pulsed mode, and contains no elements within the laser resonator cavity to reduce the intracavity power at either fundamental wavelength .lambda..sub.1 or .lambda..sub.2.